Methodology for integrated fuel lattice and fuel load optimization using population-based metaheuristics and decision trees

Abstract

A new methodology to deal in an integrated way with the fuel lattice design and fuel load optimization in a Boiling Water Reactor is proposed. The novel aspects are two. Firstly, the use of basic population based metaheuristics to deal with sets of potential fuel lattice design and sets of potential fuel reloads. Both populations evolve in a parallel and independent way between them. After some iterations the best current fuel lattice is fed to the fuel reload optimization process. Secondly, and in order to evaluate the quality of the lattice designs, the use of previously developed decision trees to speed up the calculations, thus avoiding the use of computationally expensive core reactor simulators. The computational experiments allowed to assess the benefits of our proposal. In one hand, the obtained solutions fulfilled the energy requirements and at the same time the core safety was guaranteed by thermal limits and cold shutdown margin. By the other hand, the use of decision trees allowed to speed up the process by a factor of 1200.